Fmincon's interior-point algorithm does not return minimal objective

Dear all, I am using fmincon's interior-point algorithm to find four parameters that minimize an objective function. However, fmincon does not return the minimum. I monitor the objective in each call of the objective function, and found that parameter sets exist, for which the objective is, e.g., -13.5. However, fmincon returns parameters that result in an objective of, e.g., -10.7. Is this intendend behaviour? Thanks. Kind regards, Philipp

[Torsten]

fmincon finds local minima, not necessarily global minima.

Best wishes

Torsten.

[Matt J]

What is the value of the exitflag output?

[Philipp]

The exit flag is 2.

Answer 1

[Alan Weiss]

Perhaps some of the suggestions on what to do when the solver might have succeeded will help. And, as Torsten suggested, familiarize yourself with local vs. global solutions, including how to search for a global solution.

Alan Weiss

MATLAB mathematical toolbox documentation